For B2B buyers sourcing eyeshadow palettes and powder compacts, the single most useful defect acceptance plan to write into the purchase specification is: Critical defects with zero acceptance (Ac=0/Re=1), Major defects at AQL 2.5, and Minor defects at AQL 4.0, applied under a single normal sampling plan at General Inspection Level II in accordance with ISO 2859-1 (equivalent to ANSI/ASQ Z1.4). This combination is widely used across the cosmetics packaging industry because it balances inspection cost against real-world risk: a palette with a loose pan or a failed closure is a functional failure that customers will return, while minor cosmetic deviations are tolerable within controlled limits. Defining these three levels explicitly — rather than writing a bare "AQL 2.5/4.0" — removes ambiguity about what happens when one defective unit is found in a sample of 80 pieces.
Equally important, an AQL number alone does not make a quality plan. AQL sampling is a statistical acceptance tool for visible and measurable defects at the moment of inspection; it does not test whether a magnetic closure still holds after 3,000 open-close cycles, whether the compact survives a drop onto a warehouse floor, or whether the injected plastic is compatible with the pressed powder formula. Those requirements must be specified separately, alongside a golden sample system, a defined inspection lot, and clear rules for what happens when a lot is rejected. This article walks through the defect classifications, dimensional checkpoints, sampling tables, and contractual language that buyers of eyeshadow palette packaging should put in writing before the first production order ships.
A frequent and costly contract mistake is specifying nothing more than "AQL 2.5 for Major, 4.0 for Minor" without citing the standard, the inspection level, the defect definitions, or the sampling method. Without the standard reference, the supplier may apply a special inspection level with a smaller sample size, or an accelerated single-sampling variant, and still claim compliance. Without written defect definitions, the inspector and the factory can disagree about whether a slightly scuffed mirror is Minor or Major, and that disagreement is resolved in the factory's favor on inspection day. The result is a clause that looks professional on paper but provides almost no protection in a dispute.
A complete AQL clause should contain at least five elements: the governing standard (ISO 2859-1 or ANSI/ASQ Z1.4), the inspection level (typically General Inspection Level II), the sampling plan type (single, normal), the three defect classes with written examples for this specific product, and the inspection conditions (lighting, sample source, timing relative to production completion). Buyers who add these elements turn a vague number into an enforceable acceptance criterion. The difference becomes obvious the first time a shipment contains marginal defects and the acceptance decision must be defended with an objective document rather than an opinion.
It is also worth stating in the specification what AQL does not cover. Acceptance sampling tells you the probability of passing a lot at a given defective rate; it does not guarantee that every unit in the accepted lot is defect-free, and it says nothing about durability, safety compliance documentation, or material compatibility. Treating the AQL table as the entire quality agreement is one of the most common root causes of post-shipment claims in the cosmetics packaging trade.
ISO 2859-1, which is technically equivalent to ANSI/ASQ Z1.4, is the reference standard for acceptance sampling by attributes and is the appropriate basis for visual and assembly defect inspection of eyeshadow palettes. Within the standard, General Inspection Level II is the default and provides a sensible sample size for typical packaging order quantities, from a few thousand pieces up into the hundreds of thousands. The plan type should be stated as single normal sampling, meaning one sample is drawn per lot and judged once against the acceptance and rejection numbers, rather than double or multiple sampling which complicates administration and is rarely needed for packaging.
The typical industry combination for cosmetic compacts and palettes is Critical AQL 0 (accept zero, reject on one), Major AQL 2.5, and Minor AQL 4.0. Under a single normal plan at Level II, this means that for a lot of 10,000 palettes the sample size is 125 units. Finding a single critical defect anywhere in those 125 units — a sharp burr, a contaminated interior, a wrong component — rejects the lot outright, while Major defects are tolerated up to a small count (for example, 7 in a 125-unit sample) and Minor defects up to a slightly higher count (for example, 10). The asymmetric thresholds reflect the different consequences of each defect class reaching the market.
ISO 2859-1 also contains switching rules that sophisticated buyers reference in their quality agreements. If two out of five consecutive lots are rejected on original inspection, the standard requires moving to tightened inspection, which increases sample sizes and reduces acceptance numbers. Conversely, long runs of accepted lots can justify reduced inspection. Buyers sourcing recurring orders of the same SKU should reference these rules explicitly, because they convert quality performance history into automatic consequences: a factory whose reject rate is climbing will face stricter sampling before the trend becomes a container of customer complaints.
For buyers working with new suppliers or new molds, an additional safeguard is to state that the first three production lots are always inspected under normal or tightened inspection regardless of switching-rule eligibility, and that reduced inspection requires written buyer approval. This prevents a supplier from quietly relaxing inspection intensity once the relationship is established.
Defect classification is the heart of the AQL specification, because the acceptance numbers are meaningless until each class is defined with product-specific examples. The standard three-tier structure works well for eyeshadow palette packaging, and the classes should be illustrated with concrete examples agreed against the signed golden sample rather than left to adjectives like "acceptable" and "unacceptable."
Critical defects are conditions that create safety risks, regulatory exposure, or make the product fundamentally unusable, and they carry Ac=0/Re=1: a single occurrence anywhere in the sample rejects the lot. For eyeshadow palettes and powder compacts, typical critical classifications include:
The zero-acceptance rule for critical defects is deliberate. Statistical sampling exists to balance inspection cost against defect probability, but a single sharp burr in a sample of 125 is a reliable signal that the process is producing hazards at a rate no buyer can accept, so the standard provides no tolerance at all for this class.
Major defects are functional failures that a consumer would notice and likely return, even though the product is not unsafe. At AQL 2.5 under General Inspection Level II, the acceptance numbers are tight — for example, 7 allowed in a 125-unit sample — which forces the factory to control assembly and machining processes carefully. Common major classifications for palettes include:
Minor defects are appearance deviations that do not impair function and that a reasonable consumer would accept. The AQL 4.0 threshold is more generous, but the classification must still be controlled against a reference standard, because "minor" drifts upward on every production run if it is never pinned down. Typical minor classifications include:
Two drafting rules keep minor classification honest. First, color must always be judged against the signed physical color plaque under a defined light source (D65 is the usual choice) and quantified where possible with a spectrophotometer reading of ΔE; "looks fine in the workshop" is not a judgment method. Second, the same defect type can only sit in one class — if a scratch is Major when it is visible at arm's length and Minor otherwise, that distinction must be written down with a viewing distance and angle, or the inspector will apply whichever reading favors the pass.
Defining AQL levels answers only part of the question; buyers also need to define when inspection happens. Post-production inspection alone discovers problems after the money is already spent, so mature sourcing programs layer several checkpoints through the production cycle, each with its own scope and acceptance criteria.
The First Article Inspection is conducted on units pulled from the initial production run before volume assembly proceeds, and its purpose is to verify that the tooling produces parts within the drawing. The inspector measures every critical dimension on the FAI checklist — pan pocket diameters and depths, overall height, hinge pin positions, magnet seat dimensions — records actual values against nominal values and tolerances, and the buyer signs the FAI report before mass production continues. For eyeshadow palettes, FAI should also include a trial fit of pans, mirror, and closure components from the actual production tooling, because a pocket that is 0.15 mm oversize will pass individual dimensional checks but produce loose pans in assembly. FAI is a written gate: no approval signature, no authorization to run the full quantity.
A during-production inspection, commonly called DUPRO, is performed when roughly 20–30% of the order is completed. Its value is early warning: if the sample at this stage shows systematic problems — pans fitting loosely, hinge flash, color drift from the plaque — the factory can correct tooling or process parameters while the majority of the quantity is still unproduced. DUPRO sampling does not need the full AQL formality of final inspection, but it should still be random, documented, and judged against the same defect classifications, because a "heads-up" visit without written findings creates no accountability. For first-time tooling on a new palette design, DUPRO is strongly recommended; for repeat orders on a stable, proven mold, buyers often skip it and rely on FAI plus final inspection.
The pre-shipment inspection is the formal AQL acceptance event and should be scheduled when at least 80% of the order is completed and, critically, when finished goods are packed into export cartons. Two conditions make a PSI credible. First, the sample must be drawn at random across cartons selected from the full range of completed production — not from a pallet the factory has prepared in the inspection area, which is a well-known manipulation. The inspector should select carton numbers himself from the packing list and open those specific cartons. Second, production must be at or near 100% packed for the sampled lot; inspecting at 50% completion means the remaining production is unsampled and the AQL result covers only part of the shipment. The PSI report should record sample size, defect counts by class, the applicable Ac/Re numbers, photos of representative defects, and the final pass/fail statement.
Even a passed PSI can be undermined between inspection and sailing, so buyers shipping full containers often add loading supervision. The supervisor verifies that the cartons loaded are the cartons that passed inspection — matching carton markings, quantities, and batch codes — and monitors stowage conditions, since Cosmetic Packaging can be damaged by crushing heavy cartons on top of light ones or by moisture in the container. For mixed-SKU loads, loading supervision also confirms that each SKU is loaded in the agreed quantities and orientation. It is a short, inexpensive checkpoint that closes the last gap between factory floor and destination warehouse.
Generic packaging checklists miss the mechanical features that dominate eyeshadow palette complaints, so buyers should append a product-specific checkpoint list to the inspection specification. The following items consistently account for most returns and claims in this category:
Each of these checkpoints should carry a classification decision in advance. For example, a hinge that opens fully but has noticeable lateral play might be Minor, while a hinge that binds partway is Major, and a hinge pin that falls out is Critical. Writing these thresholds into the checklist before inspection day is far more effective than debating them with the factory with goods already packed.
Dimensional requirements belong in the inspection specification as a reference table, because they give both the factory and the inspector an objective basis for judgment and they make FAI meaningful. The table below shows commonly used reference values for standard round-pan eyeshadow palettes; buyers should adjust them to their own tooling drawings and pan supplier specifications, and record actual FAI measurements against them on every new tool.
| Feature | Typical Nominal Value | Reference Tolerance |
|---|---|---|
| Pan pocket diameter (standard series) | Ø26 / Ø30 / Ø36.3 mm | ±0.05 to ±0.10 mm |
| Pan pocket depth | Per pan supplier drawing | ±0.10 mm |
| Pan-to-pocket fit clearance | Designed interference or line-to-line fit | Per assembly trial on FAI; no rocking or rotation allowed |
| Hinge alignment (lid-to-base offset) | Centered per golden sample | ≤0.3 mm |
| Overall height, closed | Per product drawing | ±0.2 mm |
| Magnet closure retention force | Per design, based on lid weight | ≥0.3 to 0.8 N depending on pan count and lid mass |
| Color difference versus signed plaque | Signed master plaque | ΔE ≤ 1.5 (measured under D65) |
Three drafting notes make this table effective in practice. First, tolerances tighter than the molding process can hold are counterproductive: specifying ±0.02 mm on an injection-molded pocket guarantees disputes, while ±0.05–0.10 mm reflects realistic tooling capability. Second, the magnet retention value must be tied to a measurement method (force gauge, pull direction, number of test units), or the number cannot be verified consistently. Third, ΔE is only meaningful when the direction of the deviation and the illuminant are fixed; a ΔE of 1.2 measured against an uncontrolled workshop light source proves nothing.
Buyers using pressed-in rather than glued pans should add a removable-pan verification to the FAI and PSI checklists, especially for refillable or magnetic palette systems. The inspection confirms that pans can be exchanged in the field without deforming the pocket, since a pocket that grips correctly on day one can still be dimensionally marginal and fail after a few refill cycles.
To make the acceptance plan concrete, buyers and suppliers should attach a sampling table derived from ISO 2859-1, General Inspection Level II, single normal inspection, to the quality agreement. The table below shows the sample sizes and acceptance/rejection numbers for common eyeshadow palette lot sizes under the recommended combination of Critical (Ac=0/Re=1), Major AQL 2.5, and Minor AQL 4.0. Both parties then read the same table during inspection, and the pass/fail decision is arithmetical rather than negotiable.
| Lot Size (units) | Sample Size Code Letter | Sample Size | Critical: Ac=0 / Re=1 | Major AQL 2.5 (Ac / Re) | Minor AQL 4.0 (Ac / Re) |
|---|---|---|---|---|---|
| 281 – 500 | F | 20 | 0 / 1 | 1 / 2 | 2 / 3 |
| 501 – 1,200 | G | 32 | 0 / 1 | 2 / 3 | 3 / 4 |
| 1,201 – 3,200 | H | 50 | 0 / 1 | 3 / 4 | 5 / 6 |
| 3,201 – 10,000 | J | 80 | 0 / 1 | 5 / 6 | 7 / 8 |
| 10,001 – 35,000 | K | 125 | 0 / 1 | 7 / 8 | 10 / 11 |
| 35,001 – 150,000 | L | 200 | 0 / 1 | 10 / 11 | 14 / 15 |
| 150,001 – 500,000 | M | 315 | 0 / 1 | 14 / 15 | 21 / 22 |
Buyers should note two practical implications of this table. First, the critical column never changes: whatever the lot size, one critical defect rejects the lot, which is why the critical class must be defined narrowly enough to be genuinely hazardous rather than a dumping ground for annoying-but-tolerable flaws. Second, the acceptance numbers for Major defects are small at typical order quantities, so a systematic assembly problem affecting even 4% of production will fail a lot of 10,000 units. Suppliers who understand this structure at order placement will invest in in-process checks; suppliers who first see the table at inspection day will treat every marginal unit as an argument.
AQL numbers operate on lots, so the specification must define what a lot is. The standard industry definition for cosmetics packaging is: one inspection lot equals the same SKU, produced with the same material batch, same color, same decoration process, and same production run. A shipment that mixes two production runs of the same SKU is two lots, each of which must meet the acceptance criteria independently, because combining them dilutes a bad run inside a good one and can allow a rejected quality level to pass on paper.
The specification should also prohibit supplier pre-selection of samples. The inspection sample must be drawn at random across the full population of packed cartons, with the inspector selecting carton numbers from the packing list. A factory that sets aside a "clean pallet" for inspection and ships the rest is not subjecting its production to the agreed statistical plan, and the contract should say so explicitly, with the consequence that any inspection performed on pre-selected goods is void.
Rejection handling should be written before it is needed. Typical contractual terms state that a rejected lot shall be reworked or replaced by the supplier at the supplier's cost, including the cost of re-inspection; that re-inspection after rework is performed at tightened severity (or at minimum, at the same severity with a fresh random sample); and that a lot rejected twice for the same defect class escalates to buyer-defined consequences, such as 100% screening of the lot at supplier cost or cancellation of the affected order. Without these terms, a failed inspection turns into a negotiation about who pays for sorting labor while the shipping deadline burns.
Finally, define the timing rule: PSI must occur at approximately 80% completion or more, with goods packed, and the inspection report issued before the agreed shipment date with enough margin for a contingency re-inspection. Buyers who allow inspections on 60%-completed, unpacked production routinely discover that the last 40% — produced after the "passed" inspection — is where the problems live.
Acceptance sampling by attributes evaluates what an inspector can see and measure in a few minutes per unit. It cannot evaluate durability or chemical behavior, so the quality agreement must list functional and laboratory tests separately, with specified methods, sample sizes, and acceptance criteria. For eyeshadow palette packaging, the usual separate-test list includes:
The frequency of these tests is a business decision: many buyers require the full functional test battery on new tooling and first production, then periodic retesting every six to twelve months or after any tooling or material change, with certificates and reports retained as part of the product file. What matters contractually is that the tests, methods, and pass criteria exist in writing — "tested to international standards" without naming the test is as empty as "AQL 2.5" without naming the sampling standard.
Palettes sold through e-commerce channels endure a rougher journey than wholesale cartons delivered to a retail distribution center: individual parcels are sorted mechanically, dropped repeatedly, stacked under variable loads, and exposed to temperature swings in last-mile vehicles. Buyers feeding this channel should tighten the quality plan in three directions. First, add parcel-level drop and stacking tests to the functional requirements, using the actual retail-ready mailer configuration rather than only the export carton. Second, raise closure expectations: lids that survive a Wholesale pallet journey may still open in a parcel-sortation environment, so magnetic retention values and snap engagement should be verified on units after drop conditioning, not only as-made.
Third, adjust the inspection emphasis. For e-commerce SKUs, buyers often shift some borderline defects upward in severity — for example, classifying a scuffed outer surface that would be Minor for wholesale as Major when the unit ships as a single consumer parcel where outer appearance is the entire first impression. Interior packaging (molded inserts, films, cushioning) also becomes part of the inspection scope, because its absence or misassembly shows up directly as transit damage rates. The incremental inspection cost is small relative to the refund, reshipment, and review costs of a damage-prone listing.
Every subjective judgment in the inspection plan — color, surface finish, decoration quality, acceptable minor deviations — is anchored by the golden sample system, and this system must be defined in the specification rather than improvised. The standard practice is a signed Pre-Production Sample (PPS): one or more units produced with mass-production tooling, materials, and processes, signed and dated by both the buyer and the factory, with the factory keeping a duplicate set. Color must additionally be controlled by signed physical color plaques, one per colorway, stored away from light and replaced on a defined schedule, because plastic and coated surfaces fade and age until they no longer represent the approved color.
The specification should state that the signed PPS and plaques are the arbitration reference for all appearance disputes, that inspection is conducted with the golden sample physically present, and that any change in material, colorant, decoration method, or tooling requires a new signed sample before the affected production can be accepted. Buyers should also specify where the reference sets live: a set at the factory for production self-checks, a set with the buyer or buying office, and optionally a set with the third-party inspection provider. When each party inspects against a different physical reference, "same quality" becomes three different things.
A final drafting detail protects both sides: record the PPS signing date and link it to the order number, and state that the golden sample supersedes any catalog images or digital proofs in case of conflict. Digital renders are marketing tools, not quality references, and factories rightly resist being held to a screen's color rendering.
Quality planning intersects with commercial terms, and buyers should evaluate MOQ, packaging, and documentation as part of the same specification rather than as separate negotiations. On MOQ: minimum order quantities exist because tooling setup, color changes, and decoration plate changes carry fixed costs, and a supplier accepting a low MOQ at marginal pricing will be tempted to recover margin by compressing quality control — exactly the inspections and in-process checks this article relies on. Buyers ordering small quantities should therefore expect and accept the cost of the full inspection framework (FAI, PSI, functional testing) as a share of order value, rather than negotiating the inspection away; a small order with no PSI is the highest-risk purchase in this category.
On packaging: export packaging is part of the inspected product. The specification should define inner packaging (individual polybags or blister trays that protect mirrors and decoration), carton gross weight limits (excessively heavy cartons drive transit crushing damage), palletization requirements, and carton marking including SKU, quantity, and production batch codes — batch codes are what make the inspection lot definition traceable in practice. Buyers should also require desiccant or moisture protection where the destination climate warrants it, since mirror delamination and coating bloom are frequently moisture-driven and typically excluded from supplier liability unless packaging was specified.
On documentation: the pre-shipment file should contain the FAI report with actual measured values, the PSI report with defect counts against the AQL table and photographic evidence, functional test reports or certificates, material and safety compliance declarations for contact surfaces, and the packing list with batch codes matched to the inspection lots. Buyers should keep these documents per order in a product file that also holds the signed PPS records and artwork approvals. Two contractual details are worth adding: first, that documents are a condition of payment release, not an afterthought; second, that material or process changes require advance written notice and, where relevant, updated compliance documentation — silent substitution of a coating or colorant supplier is a recurring root cause of both quality and regulatory problems in this product family.
Buyers assessing new suppliers should also treat documentation capability as a supplier selection criterion, not merely a compliance formality. A factory that can produce a coherent FAI report, retains batch records, and issues structured PSI reports on request is demonstrating a process culture that shows up later in consistent quality. Site visits, references from comparable cosmetic packaging orders, and a trial order with the full inspection framework applied are the practical due-diligence tools available before committing volume.
The widely used combination is Critical defects at zero acceptance (Ac=0/Re=1), Major defects at AQL 2.5, and Minor defects at AQL 4.0, applied with a single normal sampling plan at General Inspection Level II under ISO 2859-1 (equivalent to ANSI/ASQ Z1.4). Write the standard, level, plan type, and defect definitions into the specification — a bare "AQL 2.5/4.0" is not enforceable in a dispute because it does not fix the sample size or what counts as a defect.
Critical defects are defined as safety hazards, compliance failures, or conditions that make the product unusable, such as sharp burrs or contamination on contact surfaces. When such a defect appears even once in a random sample, the statistically sound conclusion is that the lot contains an unacceptable hazard rate, so ISO 2859-1 provides no acceptance number for this class. This is also why critical definitions should be kept narrow and hazard-focused.
No. AQL sampling is an attribute-based visual and dimensional acceptance method; it cannot measure retention force after thousands of cycles, survival after drops, or chemical compatibility with the powder formula. These must be specified as separate tests with defined methods, quantities, and pass criteria, typically run on new tooling and periodically thereafter, with reports from an accredited third-party laboratory where independent verification is needed.
Under General Inspection Level II, single normal sampling, a lot of 10,001–35,000 units carries a sample size code letter K and a sample of 125 units. With Major at AQL 2.5 the acceptance number is 7 (reject at 8), and with Minor at AQL 4.0 it is 10 (reject at 11), while any critical defect rejects the lot. The sample must be drawn at random across cartons selected from the full packed quantity.
The standard requirement is at least 80% of the order completed and packed into export cartons, with sampling drawn from cartons across the whole production rather than from a prepared pallet. Inspecting earlier leaves part of the order unsampled, and inspecting loose goods on a factory bench removes the carton-level checks and the protection of random carton selection. Many buyers also add loading supervision to confirm the inspected cartons are the ones shipped.
A golden sample is a unit produced with mass-production tooling and materials and signed by both buyer and supplier as the physical reference for appearance, color, and finish judgments. Signed PPS units and color plaques give inspectors an arbitration standard for subjective defects, prevent quiet drift across production runs, and should be maintained as dated, linked-to-order reference sets held by both parties. Any material, colorant, or decoration process change should require a newly signed sample before acceptance.
A defensible quality plan for eyeshadow palette packaging is a short set of documents, not a vague number: Critical defects at zero acceptance, Major at AQL 2.5, Minor at AQL 4.0, sampled single-normal at General Inspection Level II under ISO 2859-1; defect classes illustrated with product-specific examples covering contamination, sharp edges, pan fit, hinges, closures, mirrors, and decoration; a dimension and tolerance table with realistic values such as Ø26/Ø30/Ø36.3 mm pan pockets at ±0.05–0.10 mm and ΔE ≤ 1.5 against signed plaques; inspection stages at FAI, 20–30% DUPRO, PSI at 80% packed completion with cross-carton random sampling, and loading supervision; separate functional tests for closure retention, drops, cycles, and compatibility; and a signed golden sample system anchoring every subjective judgment.
Buyers who write this level of detail into the purchase specification get two compounding benefits: inspections become arithmetical decisions that factories can prepare for honestly, and disputes resolve against documents rather than negotiation leverage. Buyers who skip the detail usually discover the gaps with a container already on the water. The specification effort is modest, and in the cosmetics packaging category — where a single returned SKU listing can erase the margin of an entire order — it is the cheapest quality investment available.
Jinhua Xingqiao Plastic Industry Co., Ltd.
Mobile phone: +86-13868982062
Telephone: +86-579-82196667
Fax: +86-579-82196656
E-mail: sales@zjxqpackaging.com
Address: 377 Yongxue Street, Wucheng District, Jinhua City, Zhejiang Province, China
Copyright © 2022 Jinhua Xingqiao Plastic Industry Co., Ltd
This website uses cookies to ensure you get the best experience on our website.
Phone
Comment
（0）